Beruflich Dokumente
Kultur Dokumente
MASTER’S THESIS
Writer:
Daniel S. Jacobsen …………………………………………
(Writer’s signature)
Faculty supervisor:
Bernt S. Aadnøy
Thesis title:
Study of Slug Flow in Undulated Horizontal Wells
Credits (ECTS): 30
Key words:
Multiphase flow Pages: 92
Slug Flow
Terrain induced slugging + Enclosure: 9
Mitigation techniques
Experimental investigation
Stavanger, 15th June, 2016
Study of Slug Flow in
Undulated Horizontal Wells
Daniel S. Jacobsen
June 2016
Acknowledgements
List of figures xi
Nomenclature xv
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Scope and objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Multiphase Flow 5
2.1 Concepts of multiphase flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 Velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.2 Fluid fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.3 Pressure gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Flow regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Vertical Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 Horizontal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Flow Regime maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.1 Baker horizontal flow regime map . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.2 Mandhane horizontal flow regime map . . . . . . . . . . . . . . . . . . . . 13
2.3.3 Taitel and Dukler horizontal flow regime map . . . . . . . . . . . . . . . 14
2.3.4 Taitel, Bornea and Dukler vertical flow regime map . . . . . . . . . . . 15
3 Slug Flow 17
3.1 Slug flow related problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
viii Table of contents
4 Experimental Work 37
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2.1 Configuration 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2.2 Configuration 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2.3 Configuration 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.3.1 Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.3.2 Pipes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.4 Test conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4.1 Liquid flow rate and superficial velocity . . . . . . . . . . . . . . . . . . . 45
4.4.2 Gas flow rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4.3 Flow regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.5 Testing procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Table of contents ix
5 Results 47
5.1 Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2 Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 Experiment 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4 Experiment 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.5 Experiment 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.6 Experiment 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6 Discussion 65
7 Conclusion 69
7.1 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
References 71
Appendix A Experiments 75
A.1 Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
A.2 Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
A.3 Experiment 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
A.4 Experiment 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
A.5 Experiment 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
A.6 Experiment 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
List of figures
Introduction
1.1 Background
Slug flow is considered a major flow assurance challenge characterized by alternation
of liquid and gas flow. Large variations in flow rates and pressure is a concern for
the reservoir integrity alongside corrosion, damage to pipelines and flooding of first
stage separator. Terrain induced slugs can originate from complex well geometries,
pipeline topography and low spots in flexible risers. In addition, slugs are created from
hydrodynamic instability caused by flowing conditions. Terrain induced slugging have
become more common as the petroleum industry matures. Late-life fields, deepwater
fields and marginal subsea tiebacks to existing facilities are prone to terrain induced
slugging. Extended reach wellbore, including snake wells, fish-hook wells and undulated
wells are relative new technologies used to drain otherwise non-profitable reservoirs.
These well trajectories are, however, prone to terrain induced slugging since they can
resemble a pipeline-riser system containing low spots over large distances to accumulate
large liquid slugs.
ers, topside choking, gas injection at the riser base and subsea processing. Individual
drawbacks of these mitigation techniques are discussed in Section 3.5. In this thesis,
the potential of a device to mechanically break down or alter the slug flow behaviour is
tested in six experiments.
• Study flow in small scale flow loop and identify flow conditions for slug flow.
1.4 Outline
• Chapter 2 reviews literature of multiphase flow focusing on concepts used in
multiphase flow, flow regimes encountered in horizontal and vertical flow direction
and flow regime maps for horizontal and vertical flow.
• Chapter 3 reviews slug flow focusing on why slug flow is encountered in various
industrial applications, models describing slug flow and special emphasis on
terrain induced slugging, considered to be largest slugs encountered. Several
mitigation techniques are discussed as well as two cases from the North Sea are
reviewed.
• Chapter 4 summarizes the experimental work. Three flow loops were used to
evaluate the use of a turbine in horizontal section, inclined section and vertical
section.
• Chapter 7 concludes the work of the experimental study and suggests future work.
Chapter 2
Multiphase Flow
Multiphase flow is simultaneous flow of materials in different phases, either as gas, liquid
or solid with presence of minimum two phases. For petroleum production, presence
of oil and gas at the same time is a common multiphase flow system. This is the case
when gas-lift is used in oil wells or when the conditions are such that the produced
hydrocarbons are gas and liquid.
Conditions for a multiphase petroleum system is defined with a phase diagram. The
phase diagram illustrates conditions for a given system as function of pressure and
temperature. A two-phase flow system is the condition within the envelope in Figure 2.1.
Figure 2.1: Typical phase diagram for hydrocarbons as function of pressure and temper-
ature. Multiphase components are found within the boundary of the envelope (Terry
and Rogers, 2014)
6 Multiphase Flow
2.1.1 Velocity
The phase velocity expresses the real velocity of each phase in the flow system. Determi-
nation of the phase velocity requires knowledge of area occupied by the specific phase,
which can change throughout the system. The expression for the phase velocity is:
Qi
ui = (2.1)
Ai
As determination of the phase velocity requires detailed information about the flow
at a specific point, the superficial velocity is introduced as it only requires knowledge
about the pipe size and the volumetric flow rate of the phase. This makes the superficial
velocity recommended for multiphase flow. The superficial velocity is expressed as:
Qi
us i = = αi u i (2.2)
A
The mixture velocity is the average flow velocity and can be expressed with the superficial
velocities in the following way:
X
um = uN s i (2.3)
N
2.1 Concepts of multiphase flow 7
Void fraction
Void fraction is the fraction occupied by gas in the flow system defined geometrically ei-
ther by relative length, cross-sectional area or volume. The common method to quantify
the void fraction is with the cross-sectional void fraction (Thome, 2004) expressed as:
AG
αG = (2.4)
A
Liquid holdup
Liquid holdup is the fraction occupied by liquid in the flow system. The heavier liquid
usually flows at a lower speed than the lighter gas and is for that reason more held up,
hence liquid holdup. The liquid holdup is expressed as:
AL
αL = (2.5)
A
The void fraction and liquid holdup are linked by the fundamental relation in air-water
flow:
αL +αG = 1 (2.6)
The fractions changes along the flow due to geometrical configurations, flow regime,
pipe size and fluid properties.
dp dp dp dp
= + + (2.7)
dx dx f dx h dx a
This means that the total pressure drop is caused by friction, head loss and acceleration.
The friction term is expressed by (Thome, 2004):
dp 2
= C (Rem )−n ρm u m
2
(2.8)
dx f D
8 Multiphase Flow
Where C and n are dependent of the Reynolds number in the following way (Filip et al.,
2014):
< 2000, C = 16 n =1
Re = 2000 ≤ Re ≤ 20 000, C = 0.079 n = 0.25
≥ 20000, C = 0.046 n = 0.2
dp
= ρm g sinβ (2.10)
dx h
The inclination β is given with respect to horizontal. Furthermore, the acceleration term
is expressed as (Thome, 2004):
dp d (ṁ t /ρm )
= (2.11)
dx a dx
The model is a generalisation of a single phase flow model to a multiphase flow model
by assuming completely homogeneous flow. Other correlations have been found for
flow regimes where homogeneous flow is not the case.
Figure 2.2: Flow regimes in upward flow direction in a vertical pipe (Malekzadeh, 2012)
Bubble flow
Bubble flow is characterized by continuous liquid flow containing small gas bubbles.
Little attention has been given bubble flow as there are few industrial applications
and challenges with this type of flow. Bubble flow occurs at relatively low velocities.
Distinguishing between bubble flow and dispersed bubble flow can be a challenge,
difference being the shape of the gas bubbles. While the bubbles in dispersed flow are
spherical or nearly spherical, the gas bubbles in bubble flow are ellipsoidal or pulsating
in their shape (Andreussi et al., 1999).
Slug Flow
The gas phase flows as large bubbles separated by liquid slugs. The bubbles are bullet-
shaped, often referred to as Taylor bubbles in the literature. The gas phase travels at
higher velocities than the liquid phase resulting in liquid holdup. Both gas and liquid
phase contributes significantly to the pressure drop in this regime. Compared to bubble
flow, the velocity of the gas phase is generally higher.
10 Multiphase Flow
Churn Flow
Churn flow is considered to be the result of transition between slug and annular flow.
The liquid slugs between the gas bubbles can be discontinuous or disappear, while the
gas phase becomes continuous. The pressure drop becomes more dependent on the gas
phase, rather than the liquid phase (Bai and Bai, 2012). Compared to slug flow, churn is
more chaotic and disordered as well as gas phase velocity is increased.
Annular-mist flow
Annular flow is characterized by the gas phase flowing in the middle of the pipe with
small droplets of liquid in the stream. The rest of the liquid flows at the pipe wall as
a liquid film. This flow regime is desired because of the flow stability. Mist flow is the
regime encountered when the gas velocity becomes very high. The liquid film is thinned
by the shear of the gas, until eventually all the liquid is entrained as droplets in the
continuous gas phase (Thome, 2004).
Dispersed bubble flow is characterized by small dispersed gas bubbles moving along the
flow in otherwise continuous flow of liquid. The size of the dispersed bubbles subside
with increasing velocity of the continuous liquid phase (Bai and Bai, 2012).
Stratified flow
The gas and liquid phase in stratified flow are separated with an interface between
the phases. Smooth stratified flow is characterized with a smooth interface, whereas
stratified wavy flow is characterized by waves moving in the flow direction. Waves arise
as the result of greater gas velocity creating instabilities in the interface.
Slug flow
Elongated bubble flow contain small dispersed gas bubbles moving through a continuous
liquid phase. The flow pattern is similar to the flow pattern of slug flow, but the size of
the bubbles are generally smaller with lower velocity. Elongated bubbles are formed
when smaller bubbles coalesce, often referred to as plug flow in the literature.
Annular flow
Similar to annular flow in vertical direction, the gas phase moves along the flow direction
in the centre of the pipe with some liquid entrained as small droplets. The rest of the
liquid flows along the pipe wall as a liquid film.
in laboratories. This made for high uncertainty when used. Later developments used
data from flow data banks to develop flow regime maps, although the data sets were in
most cases results of visual observations.
ρG ρL
1/2
λ= (2.12)
ρa i r ρw a t e r
and
σ 2 1/3
µL ρw a t e r
wat e r
ψ= (2.13)
σ µw a t e r ρL
Bakers work resulted in the following flow regime map for horizontal flow:
It is important to know that the transitional zones were rather broad and the map suffered
from not having a basis in mechanisms causing transitions as the map was made from
observations by Baker.
2.3 Flow Regime maps 13
Flow pattern observations were basis for the transitions in the air-water system. The
diagram is an average compromise of the variety of combinations of pipe diameters
and physical properties. The model was better than any of the other models examined
when considering air-water data. The proposed map tends to be more accurate when
the diameter is less than 2 inches because most of the observations in the data bank
were within this range.
14 Multiphase Flow
Further, they expressed the transition from stratified to intermittent or annular dispersed
as: 1/2
ρL −ρG g cosβ A G
hL
u G = 1− (2.15)
D ρG Si
Distinguishing between transition intermittent flow and annular dispersed flow was
found to be affected by the hL /D -ratio. In the paper it was suggested transition to
intermittent flow when hL /D < 0.5, otherwise transition to annular dispersed flow. A
modified criterion of hL /D < 0.35 was later suggested to account for gas holdup in the
liquid slug (Barnea et al., 1982). Further, they found transition between intermittent
flow and dispersed bubble flow to happen when:
1/2
4A G g cosβ ρL −ρG
uL > (2.16)
Si ρL f L
Figure 2.6 illustrates the resulting flow regime map from the study, including a compari-
son with the Mandhane plot from Figure 2.5, indicating good agreement between the
flow regime maps.
2.3 Flow Regime maps 15
Figure 2.6: Taitel and Dukler horizontal flow regime map for a 2.5 cm inner diameter
pipe, air-water flow at atmospheric conditions. For comparison, the Mandhane plot is
added (Taitel and Dukler, 1976)
It is seen by the statement that bubble flow is heavily dependent on the diameter of the
pipe, and if the pipe diameter is sufficiently high, bubble flow is not existing. Further,
they studied the mechanism that caused transition from bubble flow to slug flow and
found the following expression to be fulfilled for bubble to slug transition:
1/4
g (ρL −ρG )σ
u S L = 3.0u SG −1.15 (2.18)
ρL2
16 Multiphase Flow
Further, they found that the transition from bubble flow to dispersed bubble flow was
expressed with the following equation:
Transition from slug flow to churn flow was found to be expressed as:
lE um
= 40.6 p +0.22 (2.20)
D gD
1/2
u SG ρG
1/4 = 3.1 (2.21)
g (ρL −ρG )σ
Equation 2.21 shows that annular flow is independent of the liquid flow rate and pipe
diameter. The equations above was used to make the flow regime map illustrated in
Figure 2.7.
Figure 2.7: Bornea, Taitel and Dukler vertical flow regime map for a 5.0 cm inner diameter
pipe, air-water flow at atmospheric conditions, σ = 10N/cm2 and varying l E /D for Slug-
Churn transition (Taitel et al., 1980)
Chapter 3
Slug Flow
Slug flow is a multiphase flow regime characterized by alternating flow of gas and liquid
slugs. Slug flow in the well and pipelines are undesirable because of large fluctuations in
both pressure and flow rates, ultimately leading to decrease in the overall production.
As the behaviour of slug flow is complex in nature, accurate predictions are challenging,
especially as there are several parameters affecting the flow behaviour. Slug flow in wells
and pipelines are common because of hydrodynamic instability, complex well geometry,
topography and flexible riser configurations. As the current mitigation techniques reduce
overall production rates or require additional equipment taking up large spaces, new
mitigation techniques are desired. Developments of deepwater and marginal subsea
tiebacks increase the likelihood of slug flow as the terrain becomes more complex and
the distance to processing facility increases.
Figure 3.2: Slug unit cell for horizontal flow, l u is the length of one unit cell, l f is the
length of the liquid film, l s is the length of the liquid slug and l M is the mixing zone
within the liquid slug (Dukler and Hubbard, 1975)
The slug unit is divided into two parts, the liquid slug part, also called the the slug body
with length l s and the liquid film with length l f . It is seen that the length of one unit is
expressed as:
l u = l f +l s (3.2)
Figure 3.3: Idealized pressure drop in slug flow utilizing the concept of unit cell (Dukler
and Hubbard, 1975)
Two contributions to the pressure drop across the slug is observed. ∆Ps is the friction
in the liquid slug from overcoming wall shear in the back of the slug. ∆PM is pressure
drop due to acceleration of the liquid film to slug velocity in the mixing zone. The total
20 Slug Flow
pressure drop over the slug is expressed as (Dukler and Hubbard, 1975):
Shell Slug Frequency Correlation (Stapelberg and Mewes, 1994) was derived by curve-
fitting the data of Heywood and Richardson, getting the following relation for the slug
frequency:
The correlation was found to give good agreement with measured data for a 4" air-water
inclined flow loop. Zabaras model (Zabaras et al., 1999) represented an extension of the
Gregory and Scott correlation, implementing inclination angle from horizontal. The
correlation by Zabaras resulted in the following slug frequency equation:
1.2
u S L 19.75
0.836+2.75sin0.25 β
f s = 0.0226 + um (3.7)
gD um
Manolis et al. (Manolis et al., 1995) collected data for air-water flow at various pressures
using the approach adopted by Gregory and Scott and proposed the following correlation
for slug frequency:
1.8
i n + um
2 2
uS L u m,m
f s = 0.0037 2
(3.8)
gD um
Figure 3.4: Slug frequency as function of superficial liquid velocity for discussed models
i n = 15.5m s SG = 3.1831ms
2 2 −2 and u −1
with a 4.0 cm inner diameter pipe, u m,m
Gokcal et al. (Gokcal et al., 2009) conducted slug frequency experiments with high
viscosity oil ranging from 0.181 to 0.589 Pa·s and found that the slug frequency was
significantly affected by the liquid viscosity. This is a drawback with the models presented
here as none of them includes the liquid viscosity in addition to other influencing
parameters.
22 Slug Flow
Figure 3.5: Variations of pressure and flow rates in terrain induced slug cycle
Fish-hook wells have well geometry like a fish-hook, drilled downwards followed by an
uphill trajectory. Low spots are created where liquid slugs can accumulate.
Figure 3.6: Application of fish-hook well geometry. First drilled reservoir section located
deeper than the reservoirs drilled at the end (Malekzadeh, 2012)
24 Slug Flow
Fish-hook wells drain hydrocarbon zones located shallower and a distance away from
the reservoir first drilled through. Marginal hydrocarbon zones can be exploited to
enhance the overall production from a field. Low spots are naturally generated because
the well is drilled upwards and liquid accumulation can take place.
Undulations
Undulations create low spots for slugs to accumulate. When stratified flow is encountered
in the downward flow direction, slug flow can be present in the upward flow direction.
The same effect is seen in pipelines as the topography changes.
Figure 3.7: Slugging due to undulations in the well, flow from left to right (Feesa, 2003)
Undulations are also found in snake wells. These wells are characterized by drainage
from several vertically stacked layers with a well trajectory that is laterally weaving to
reach all the zones (Obendrauf et al., 2006). The benefit is drainage from several zones
with lower costs than multilateral wells. Geosteering utilizes logging tools to navigate
horizontal layers in the reservoir. Optimal placement of the well can result in undulations
with low spots for slug accumulation. Horizontal wells can have undulations caused by
disturbances while drilling the horizontal.
Hydraulic fractured horizontal shale oil wells with extremely low permeability and pro-
ductivity index are prone to slugging. Since the drainage radius is limited, the reservoir
pressure will deplete rapidly below the bubble point and multiphase flow is the result.
Fracturing techniques used in horizontal shale wells require liners to be of a certain
size, normally 4 or 6 inches, thereby resulting in low flow velocity within the pipe and
unstable flow (Norris, 2012).
Since the well path is rarely truly horizontal, it is reasonable to assume some small
inclinations from the horizontal in long reservoir sections making low spots for liquid
3.4 Terrain induced slugging 25
accumulation. This was studied by H. Lee Norris (Norris, 2012) by performing simula-
tions on a typical hydraulic fractured shale well with toe-up of +0.5°, creating a low spot
at the heel. The results were periodic liquid production and fluctuations as expected
from terrain induced slug flow. The slugging cycle was found to be long because of low
gas production rate and long horizontal section, thus the pressure build-up by the gas
was slow.
1. Slug formation
3. Blowout
4. Liquid fallback
Figure 3.8: Flexible riser configurations with low spots also by the shape of the riser
configurations (Jahanshahi, 2013)
The liquid level in the riser increases as both phases continue to flow into the pipeline
while the gas passage is blocked resulting in pressure increase at the riser base, push-
26 Slug Flow
ing the gas-liquid interface in the pipeline even further away from the riser base and
compressing the gas in the pipeline. The liquid slug grows larger within the riser. The
pressure at the riser base reaches its maximum when the liquid level reaches the riser
top, and the pressure of the gas in the pipeline eventually becomes higher than the
hydrostatic head of the of the liquid in the riser. Liquid starts to flow out at the top of
the riser while the slug tail pushes towards the riser base. When gas enters the riser, the
hydrostatic head in the riser decreases, the gas expands and the liquid column flushes
out of the riser. The gas flows through the riser and the liquid slug is produced. The gas
in the riser is produced rapidly, causing quick de-pressurization of the system. When all
the gas is produced, the pressure reaches its minimum. The cycle is finished and new
blockage can yet again take place at the riser base to start a new slug cycle (Malekzadeh
et al., 2012). Figure 3.5 illustrates the typical flow rate and pressure variations in a slug
cycle of this type.
The Bøe criterion (Bøe, 1981) is based on force balance applied to the liquid slug blocking
the entrance into the riser. The considered forces are the pressure build-up of gas as it is
blocked from entering the riser, seen as compressed gas and the hydrostatic head of the
3.4 Terrain induced slugging 27
liquid inside the riser. The Bøe criterion is given by the following equation (Bøe, 1981):
Pp
uS L ≥ u SG (3.9)
ρL g αG L
or as
ρG 0 RT
uS L ≥ u SG (3.10)
ρL g αG L
When the statement is valid, severe slugging can occur. Drawing from Equation 3.9 and
Equation 3.10, we see that the chance of severe slugging to occur is reduced by adjusting
key variables in a beneficial way;
• Decreasing superficial liquid velocity, liquid density, average void fraction and
pipeline length
The stability criterion by Taitel (Taitel, 1986) is based on force balance. Severe slugging
occurs because gas is compressed until it overcomes the gravitational head of the liquid
in the riser resulting in a long liquid slug that is pushed in front as the gas enters the
upstream riser and expands. Assume that the slug tail has just entered the riser and the
riser is now filled with liquid and a small disturbance y may carry the liquid somewhat
higher. The disturbance is fast enough to not affect the flow rates of liquid and gas (Taitel,
1986). This condition is illustrated in Figure 3.9.
Drawing from Figure 3.9, the net force per unit acting on the liquid film in the riser is
expressed as (Taitel, 1986):
αG L
∆F = Ps e p +ρL g h ′ − Ps e p +ρL g h − y (3.11)
αG L +αG y
If ∆F increases with y , the liquid column will be blown out of the pipe. Thus, the
condition for stability is satisfied for:
∂ (∆F )
< 0, wheny = 0 (3.12)
∂y
Applying the stability condition to the net force in the riser, the criterion for stability is
expressed as:
′
Ps e p (αG /αG )L −h
> (3.13)
P0 P0 /ρL g
′
Here, αG is the gas holdup and αG is the gas holdup in the gas cap penetrating the liquid
column. Drawing from Equation 3.13, we see that the stability of severe slugging can be
altered by adjusting key variables in a beneficial way;
• Increasing separator pressure, height of the riser, liquid density and gas holdup in
the gas cape penetrating the liquid column.
Pots criterion
The criterion by Pots et al. (Pots et al., 1987) is based on force balance where the rate of
the hydrostatic pressure build-up in the riser must exceed the pressure build-up rate
of the gas in the pipeline. With these conditions satisfied, the liquid fills the riser faster
than gas pressure drives the flow.
z RT /M wg
Πs s = (3.14)
g L αG w L
Equation 3.14 expresses the ratio between the pressure build-ups and severe slugging
can occur when Πs s < 1. Drawing from Equation 3.14, we see that the severe slugging
can be avoided by adjusting key variables in a beneficial way;
• Decreasing pipeline length, average gas holdup in the pipeline and liquid mass
flow rate
3.4 Terrain induced slugging 29
The model proposed by Jansen et al. (Jansen et al., 1996) includes elimination of severe
slugging to the criterion proposed by Taitel (Taitel, 1986) as back pressure increases
with implementation of choking at the riser top. They assumed that the two-phase time
averaged pressure drop across the choke could be approximated by (Malekzadeh, 2012):
∆Pc h o k e = C u S2 L (3.15)
The increase in pressure upstream of the choke was written as (Jansen et al., 1996):
PB − Ps e p +C u S2 L = K y
(3.16)
The net force on the interface between the end of the liquid slug and the front of the
penetrating gas phase was expressed as (Jansen et al., 1996):
αG L
∆F = Ps e p +C u S2 L +ρL g h
′
αG L +αG L
− Ps e p +C u S2 L + K y +ρL g h − y
(3.17)
The left part of the right hand side represents expansion of gas while the right part is the
resulting back pressure caused by liquid column(h − y ), separator pressure and choking.
Utilizing that severe slugging is not possible when
∂ (∆F )
< 0, when y =0
∂y
αG L
Ps e p +C u S2 L ′ 1− ρKL g −h
αG
≥ P0
(3.18)
P0
ρL g
When there is no methods of eliminating severe slugging, the criterion overlaps the Bøe
criterion for severe slugging at the top as seen in Figure 3.10a. Drawing from the stability
criterion in Equation 3.18, we see that the severe slugging can be avoided by adjusting
key variables in a beneficial way;
30 Slug Flow
• Decreasing gas holdup in the gas cape penetrating the liquid column and liquid
density
Figure 3.10: Choking effect on severe slugging. It is seen that choking reduce the enve-
lope where severe slugging can occur (Jansen et al., 1996)
It was identified in the work of Yula Tang et al. (Tang et al., 2007) and the work of
Malekzadeh and Mudde (Malekzadeh and Mudde, 2012) that severe slugging could occur
in wells caused by undulations and complex well trajectories by performing dynamic
wellbore simulations using OLGA. The models mentioned in this section assumes flow
going from downward inclined to vertical as seen in pipeline-riser systems, but this is
not the case for reservoir sections and the inclination should be included. Seen from
Equation 2.10, the inclination is affecting the hydrostatic pressure drop. Further, the Bøe
criterion of Equation 3.9 can be modified to include the inclination as follows (Ogazi,
2011):
Pp
uS L ≥ u SG (3.19)
ρL g αG sinβ L
Where the inclination is given with respect to horizontal for the upward flow direction.
measures can utilize the choking possibilities with an active controller to suppress the
slugs.
Figure 3.11: Pressure drop in a riser with choke as mitigation technique for severe
slugging with u S L = 2.0ft/s. It is seen that Riser + Choke reduces the superficial gas
velocity needed to obtain stable flow, opposed to not having a choke (Schmidt et al.,
1985)
Assume a flow rate in the region of stable flow. An increase in gas flow rate results in in-
creased pressure drop in the riser, which in turn requires higher pipeline pressure. Higher
pipeline pressure can only be achieved by reducing the flow rate out of the pipeline since
the inflow is assumed constant (Schmidt et al., 1985). Studying the statement above with
the Bøe criterion from Equation 3.9, we see that increasing pipeline pressure increases
the right hand side of the Bøe criterion, thus reducing the likelihood of severe slugging
in the riser.
requiring less degree of choking. The choking stabilizes the flow by increasing the liquid
velocity while the gas injection stabilizes the flow by increasing the gas velocity (Jansen
et al., 1994). Cost savings can be achieved because less available gas are required.
3.5.5 Controllers
Dynamic measurements from the well or production system is fed to controller systems
that stabilize the flow and suppress the slugs. Correct field data and models are crucial for
the controllers to function as intended to ensure optimized production. Main drawback
with controllers is the lack of robustness as the system becomes unstable after some time
when operating conditions change. The operators, instead of tuning the controller again,
often switch over to manual choking when the controller becomes unstable (Jahanshahi,
2013). The control system is dependent on the operating conditions of each field and
must be tuned in order operate correctly for a specific field and operating conditions.
PI-controller and PID-controller uses the feedback from monitoring to adjust the
flow conditions to obtain stable flow. A PID-controller has the advantage of reducing the
oscillations compared to PI-control and process response time can be reduced. On the
other hand, tuning is harder than for PI-control as there are more parameters involved.
A Shell development (Yaw et al., 2014) for handling slugs in pipeline-riser systems
is the Smart Choke based on a single control valve installed between the riser top and
first stage separator. The Smart Choke is compact and cost efficient compared to other
measures to mitigate slug flow. Pressure readings are used as information about incom-
ing flow which then are used in the control algorithm for the Smart Choke. The control
mechanism aims to maintain constant volumetric flow rate at the outlet via a fast acting
flow controller.
Figure 3.12: Subsea processing with separation of gas and liquid on the seafloor before
transportation to processing facility utilizing two flowlines (Haheim et al., 2009)
Although pipeline-riser slugging can be avoided by utilizing two flowlines, slugging from
the wells are not avoided. Another benefit from subsea processing is the possibility of
boosting to increase the pressure in the pipeline, making it possible to produce from
low-pressure reservoirs. The increased pipeline pressure increases the right hand side
of the Bøe criterion from Equation 3.9, thus reducing the possibility of severe slugging
in the pipeline-riser system. Subsea processing can also increase the recovery rate from
a field.
as artificial lift. The topography caused terrain induced slugging in the pipeline 8-9km
from the subsea template as seen in Figure 3.13.
Figure 3.13: Yme B pipeline topography causing terrain induced slugging 8-9 km from
the Yme B subsea template (Øverland and Ramstad, 2001)
Topside choking was implemented to reduce the heavy slugging at the expense of
production rates. The slug catcher at the processing facility was not designed to handle
heavy slugging and did not reduce the pressure fluctuations and production rate peaks
sufficiently to avoid shut downs. Production losses were significant as all the wells were
affected in a shut down scenario. Upgrade of the slug catcher was considered, but was
not found economically acceptable due to the production losses during installation
period and the high cost. This aspect shows the importance of designing the slug catcher
correctly and how it must be able to handle the largest anticipated slugs in the production
system. The instabilities increased with increasing gas lift rate as seen in Figure 3.14.
Figure 3.14: Effect of gas lift rate on slug length from Yme Beta Øst. (a ) Gas lift rate of
150 kSm3 /d, lower frequency and slug length. (b ) Gas lift rate of 250 kSm3 /d, higher
frequency and slug length (Øverland and Ramstad, 2001)
36 Slug Flow
Three possibilities of terrain induced slugging was identified, from the S-shaped riser,
the riser base and the well. It was identified that the heavy slugging originated from the
well with long slugging cycle and high pressure variations.
Decision was made that active controllers should be utilized to stabilize the flow using
feedback from downhole pressure measurements. The topside choke was controlled
by the pressure measurements in the well. The flow was effectively stabilized, but the
topside choke only had the possibility of controlling one well. Use of the subsea choke
instead of the topside choke was proposed and was found to stabilize the flow oscillations
from the well effectively. The control system for the subsea choke used pressure reading
downhole, just as for the topside choke.
Chapter 4
Experimental Work
4.1 Introduction
Experiments have been conducted in a 40 mm inner diameter pipe with air-water flow.
The experiments have been conducted to evaluate the potential use of a rotating device
with propeller and mixer to alter the flow behaviour in the pipe in a significant and
beneficial way. Three geometrical configurations have been made to study the impacts
of placing the turbine in horizontal, vertical and inclined sections.
4.2.1 Configuration 1
This configuration was made to simulate the geometry of undulations to study terrain
induced slugging. The problem was short distance of the bend sections and dimensions
of the flow loop, not allowing pressure build-up and continuously transported gas. Slug
formation happened in the loop, but the characteristics were small slug lengths with low
frequency. Chaotic behaviour was observed at the base of each bend section.
38 Experimental Work
Figure 4.1: Schematic of flow loop 1 consisting of two bend sections to simulate the
geometry of undulations in a well section or pipeline topography, lengths given in mm.
Using the modified Bøe criterion of Equation 3.19 and assuming standard conditions,
air-water flow, and u S L = u SG , it was found that the minimum length for terrain induced
slugging to take place was approximately 25 m for this geometrical configuration.
4.2 Experimental setup 39
4.2.2 Configuration 2
Configuration 2 was made because of drawbacks with configuration 1 as it did not create
large slugs that the turbine could break down, but instead continuously transported gas
as small bubbles. The same problems took place in configuration 2 as pressure build-ups
were not possible and gas was continuously transported along the flow. By studying the
loop with the Bøe criterion of Equation 3.9, it was seen that the bend section required at
least 12-13 m length to accumulate terrain induced liquid slugs at the base.
Figure 4.3: Schematic of flow loop 2 consisting of a longer bend section, lengths given
in mm.
Figure 4.5: Illustration of bend section flow in configuration 2, positions (A), (B) and (C)
explained below
• (A) Flow in the horizontal section. Liquid phase velocity is here lower than at (B),
but the liquid is accelerated close to the bend section due to gravity.
• (B) Liquid flow in inclined section is further accelerated by the gravitational pull,
increasing the liquid phase velocity.
• (C) Chaotic zone where small gas bubbles are dispersed in the water zone. The
gas is dragged with the flow because of the interaction between the phases at
the interface and some of the gas is dragged with the flow to the vertical section,
thereby continuously transporting some gas.
(c) High gas flow rate (d) Very high gas flow rate
From Figure 4.5 and Figure 4.6, it was seen that gas was transported efficiently with the
liquid flow, thus not blocking the free gas passage and no terrain induced slugging took
place.
4.2.3 Configuration 3
Configuration 3 was made to study hydrodynamic slugging as described in Section 3.2.
Hydrodynamic slugs quickly occurred close to the inlet and grew in size along the flow.
The nature of hydrodynamic slugs differs a lot from the nature of the terrain induced
slugging
Figure 4.7: Schematic of flow loop 3 consisting of horizontal section to study the effect
of hydrodynamic slugging, lengths given in mm.
4.3 Equipment
4.3.1 Turbine
Propeller
Two propellers were considered for the experiments consisting of three and six blades,
respectively. The six bladed propeller was found to function best in the experiments
for this thesis because the impulse from the water was the driving mechanism that
resulted in rotation. The function of the propeller was to rotate the turbine and it was
seen through testing that the water had the biggest impact.
Bearings
Originally the ball bearings caused high friction because they were fitted with grease.
The turbine was taken apart, the grease were removed and the bearings were sprayed
with Teflon to reduce the friction. Before the modification to the bearings, the turbine
was not rotating in horizontal flow flow direction. The bearings were the connection
part between the housing and the rotating shaft and used to reduce friction compared
to having the shaft rubbing against the housing.
Mixing part
The mixing part of the turbine was used to disperse the gas in the slug flow and affect
the gas-liquid interface by mixing the phases together. The mixing part was essentially
the part that was meant to mechanically break down the slugs.
Housing
The housing part was the connection point between the turbine and the pipes used to
lock the turbine in place with a screw. It was found that the surface area occupied by
the housing was approximately 470 mm2 . Further, the reduction of flow area over the
turbine because of the housing was approximately 37.5 %.
Figure 4.13: Illustration of surface area occupied by the turbine in the flow systems
4.3.2 Pipes
Transparent acrylic glass pipes Ø50x40 mm were used in the experiments. In Figure 4.14
two different connections are used. The pipe at the bottom had the connections used
in experiment 1 to 5, but for experiment 6 it was decided to use white union muffs to
connect each pipe because the other connections seemed to affect the flow behaviour
with small local disturbances. Bend sections and the original connections were glued
together using acrifix.
4.4 Test conditions 45
(a) PVC union muff with O-ring to seal (b) White PVC double muff
Figure 4.15: Pipe connections used in the experiments. (a) Connection in experiment
1-5 (b) Connection in experiment 6 (Dahl, 2015a,b)
20l
QL = = 0.25ls−1 (4.1)
79.5s
Further, the superficial liquid velocity was calculated by using Equation 2.2 as follows:
Q L 0.25·10−3 m3 s−1
uS L = = π 2
= 0.20ms−1 (4.2)
4 (0.04m)
A
The low superficial liquid velocity limited the range of the experimental study substan-
tially. However, it is important to identify that slug flow was expected in flow loop 3 with
horizontal flow according to the flow regime maps from Figure 2.6. The liquid flow rate
was kept constant at the maximum rate in the experiments.
individual experiment was carried out within a short time span. The gas flow rate did
not cause any restrictions for the conducted experiments. In the experiments, the gas
flow rate was high as shown in Figure 4.16.
Figure 4.16: High gas flow rate close to the inlet as used in the conducted experiments.
• Configuration to be studied was set up, all sections could easily be changed.
• The water rate was set to the maximum rate of Q L = 0.25ls−1 .
• The gas rate was set to desired value to get slug flow in configured loop. Between
each test, a valve was used as on/off-controller for the system to keep the gas rate
constant.
• Camera was used to capture effects from the turbine and a yardstick was attached
to the pipe to measure slug lengths (l s ).
• A series of tests were conducted at once to ensure same conditions for each exper-
iment.
Chapter 5
Results
Several experiments were conducted to test the placement of the turbine and the effect
on the flow behaviour.
• Experiment 3: Modified horizontal flow loop with one bend section and turbine in
downward inclined section
5.1 Experiment 1
The horizontal flow loop configuration was used with the turbine placed 120 cm from
the inlet. First test was conducted with the turbine 120 cm from the inlet and captured
on film. Further, the turbine was taken out and results captured on film. Flow conditions
were u S L = 0.20ms−1 and high gas rate. Test interval was 70 to 170 cm before the vertical
section. The distance from the inlet to the turbine is measured from the inlet to the
starting point of the housing.
0.8
Without turbine
0.7 120 cm from inlet
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.1: Size distributions of liquid slugs from placing the turbine 120 cm from inlet
in horizontal direction
The results indicated an impact from placing the turbine close to the inlet in the horizon-
tal flow direction. The frequency increased from 0.408 Hz to 0.617 Hz, but the average
slug length was reduced. Even though the total frequency increased, the frequency
for l s > 10 D decreased from 0.300 Hz to 0.171 Hz with the turbine. Slugs with size
l s = 15 − 20 D decreased by 83 %. It was also seen that slugs with l s > 20 D with tur-
bine 120 cm from the inlet was 0 while there was 29 slugs without the turbine in the flow
system.
5.2 Experiment 2 49
5.2 Experiment 2
The horizontal flow loop configuration was used with the turbine placed 195 cm from
the inlet. First test was conducted without the turbine in the system. Further the turbine
was placed in the flow loop and second test was conducted. Flow conditions were
u S L = 0.20ms−1 and high gas rate. Test interval was 70 to 170 cm before the vertical
section.
0.8
Without turbine
0.7 Turbine 195 cm from inlet
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.2: Size distributions of liquid slugs from placing the turbine 195 cm from inlet
in horizontal direction
The results indicated an impact when the turbine was placed 195 cm from the inlet
in horizontal flow direction. The results without the turbine in the loop was different
from the curve from experiment 1. The frequency increased from 0.429 Hz to 0.538 Hz.
Like in experiment 1, the frequency for l s > 10 D decreased in experiment 2 as well,
from 0.325 Hz to 0.138 Hz with the turbine 195 cm from the inlet. Slugs with size of
l s = 15−20 D decreased by 79 %. It was seen that slugs with l s > 20 D with turbine 195 cm
from the inlet was 0 while 10 were observed without the turbine.
50 Results
0.8
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.3: Comparison between 120 cm from inlet and 195 cm from inlet
The comparison between the turbine 120 cm and 195 cm from the inlet illustrated that
the slug sizing with the turbine in the system had similar trends, but this was not the
case without the turbine in the flow loops. The gas rate was not kept constant because
the tests were conducted at two separate days which might have led to some differences
between the two plots, but the largest impact was believed to be caused by changing
the pipe positioning to allow for the turbine to placed 195 cm from the inlet. The pipe
connections were changed for experiment 6 as described in Section 4.3.2.
5.3 Experiment 3 51
5.3 Experiment 3
The horizontal flow loop was modified to have one bend section in the middle. The
turbine was placed in the inclined section in the downward flow direction with the
mixing part close to the base point of the bend section. The first test was conducted
with the turbine and the second test was conducted without. Flow conditions were
u S L = 0.20ms−1 and high gas rate. The test interval was closer to the vertical section
because of short distance from the bend section to the original test interval. The results
indicated no significant impacts from placing the turbine in flow loop with one bend
section. No slugs of size l s > 20 D were observed for both tests.
0.8
Without turbine
0.7 With turbine
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.4: Slug lengths from placing turbine in bend section in modified flow loop
containing one bend
52 Results
5.4 Experiment 4
The configuration of flow loop 1 was used with two turbines, one in each bend section.
The first test was conducted without the turbines and the second test was conducted
with. Flow conditions were u S L = 0.20ms−1 and high gas rate. The test interval for this
experiment was at the top horizontal section because of short length from second bend
section to vertical section, 90 to 190 cm from the vertical section. The results indicated
no significant impacts from placing two turbines in flow loop with two bend sections.
No slugs of size l s > 20 D were observed for both tests. Compared to Experiment 3, which
had one bend section, the frequency was drastically reduced with two bend sections,
probably caused by the gas accumulating at the top between the bend sections.
0.8
Without turbine
0.7 With turbine
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.5: Slug lengths from placing turbines in bend sections of flow loop 1
Figure 5.6 illustrates how the gas was efficiently transported with the flow in the case
of flow in bend section. No difference between the two cases was seen supporting the
results from experiment 3 and 4.
5.5 Experiment 5 53
t = 0.00s t = 0.00s
t = 0.10s t = 0.10s
t = 0.20s t = 0.20s
t = 0.30s t = 0.30s
Figure 5.6: Comparison with and without turbine in bend section. Left: with turbine,
right: without turbine
5.5 Experiment 5
The horizontal flow loop was used in this experiment. The turbine was placed in the
vertical section with the propeller at the top and mixing part at the bottom. The first
test was conducted without the turbine and the second test was conducted with. Flow
conditions were u S L = 0.20ms−1 and high gas rate. The test interval was at the top
horizontal section, 90 to 190 cm from the vertical section. The results indicated minor
impacts from placing the turbine in the vertical section of the horizontal flow loop. The
frequency decreased from 0.321 Hz to 0.313 Hz, considered insignificant. Frequency for
l s > 10 D decreased from 0.300 Hz to 0.171 Hz. Slugs with size of l s = 15−20 D decreased
by 29 % and slugs with size of l s > 20 D decreased by 19 %.
54 Results
0.8
Without turbine
0.7 With turbine
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.7: Slug lengths from placing turbine in vertical section of the horizontal flow
loop
Figure 5.8: Breakdown of small gas bubble in vertical upward flow direction, u S L =
0.20ms−1 and low gas rate
5.5 Experiment 5 55
Figure 5.9: Slug flow over turbine in vertical flow direction, u S L = 0.20ms−1 and medium
high gas rate
It was seen from Figure 5.8 that the turbine was able to break down small gas bubbles in
the vertical flow direction, but when Taylor bubbles ascended over the turbine, it was
not able to break the gas bubble, as shown in Figure 5.9. The turbine stopped as the
Taylor bubbles flowed across the turbine.
56 Results
5.6 Experiment 6
The horizontal flow loop from previous experiments were modified slightly by changing
the connections to remove small disturbances caused by the original connections. First
test was conducted without turbine, followed by tests with the turbine 80, 100, 120, 140,
160 and 200 cm from the inlet. The last test was conducted without the turbine in the
flow loop. Flow conditions were u S L = 0.20ms−1 and high gas rate. Test interval was 70
to 170 cm before the vertical section.
It was seen from Figure 5.10 some differences for the values l s = 10−15D and l s =
15−20D for the two cases without the turbine. The difference was small enough to be
caused by the randomness in slug lengths for hydrodynamic slugs. The plot indicated
that large slugs were occurring without the turbine in the flow system.
0.8
Test 8 - Without
0.7 Test 1 - Without
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Further it was seen from Figure 5.11 impacts of placing the turbine in various positions
from the inlet. The plot indicated that the amount of larger slugs decreased as the turbine
was placed closer to the test interval. This could be caused by the distance from the
turbine to the test interval being shorter and not enough distance for the slugs to grow. It
was identified, however, that all the trends with the turbine had reduced relative number
5.6 Experiment 6 57
of large slugs, while without the turbine the relative number of slugs were highest for
slugs with size l s > 20 D.
0.8
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.11: Impact of turbine on slug sizes with the turbine at various positions from
the inlet
It was seen by the comparisons in Figure 5.12 and Figure 5.13 that the turbine reduced
the slug length effectively over short distances. The results revealed that the number
of slugs with l s > 20 D with the turbine in the system were 11 over six tests, while 79
slugs with l s > 20 D were recorded in the two tests without the turbine in the system.
The average overall frequency increased with the turbine from 0.360 Hz to 0.481 Hz,
while the average frequency when excluding the slugs with size l s < 10 D decreased from
0.308 Hz to 0.177 Hz with the turbine. Slugs with size of l s = 15−20 D decreased by 31 %
and slugs with size of l s > 20 D decreased by 95 %.
58 Results
0.8
Test 1 (without) Test 2(80 cm)
0.7 Test 3(100 cm) Test 4(120 cm)
Test 5(140 cm) Test 6 (160 cm)
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
0.8
0.6
Relative number of slugs
0.5
0.4
0.3
0.2
0.1
0
<10 10-15 15-20 >20
Slug length(l s /D )
Figure 5.13: Comparison of average values with and without turbine from experiment 6
5.6 Experiment 6 59
0.9
> 20 (l s /D ) 15−20 (l s /D )
0.8
10−15 (l s /D ) < 10 (l s /D )
0.7
Relative number of slugs
0.6
0.5
0.4
0.3
0.2
0.1
0
80 100 120 140 160 200
Figure 5.14: Effect of turbine placement on slug sizes and relative number of slugs
Seen from Figure 5.14 above, slugs with size l s > 20 D were not observed for placement
140-200 cm from the inlet. For slugs with size l s = 15−20 D, it was seen that the relative
number of slugs decreased as the turbine was placed longer distance from the inlet. The
opposite was the case for slugs with l s < 10 D as the relative number increased as the
turbine was place longer distance from the inlet.
From Figure 5.15 on the next page, it was seen that the liquid height increased close to
the propeller which seemed to accumulate a large number of slugs, possibly explaining
the increased frequency. Further, from Figure 5.16, as the liquid slug flowed across the
turbine, it was seen from t = 0.25s and t = 0.375s that cavitation happened, probably
caused by Bernoulli effect as the flow area decreased.
60 Results
t = 0.00s
t = 0.10s
t = 0.20s
t = 0.30s
t = 0.40s
t = 0.50s
t = 0.60s
t = 0.70s
t = 0.80s
Figure 5.15: Formation of small slug over turbine in horizontal direction with rotation,
u S L = 0.20ms−1 and medium gas rate
5.6 Experiment 6 61
t = 0.00s
t = 0.125s
t = 0.25s
t = 0.375s
t = 0.50s
t = 0.625s
Figure 5.16: Slug flow over turbine in horizontal flow direction, u S L = 0.20ms−1 and
high gas rate
62 Results
5.7 Summary
To summarize the experimental results, key findings are listed as bullet points below
and in Table 5.1;
Table 5.1: Test results summarized from all experiments. Extended information in
Appendix A
With turbine 96 28 5 0
Experiment 2
Without turbine 25 44 24 10
With turbine 77 43 23 0
Experiment 3
Without turbine 80 37 24 0
With turbine 55 24 3 0
Experiment 4
Without turbine 51 23 2 0
With turbine 26 22 16 11
Experiment 5
Without turbine 16 24 23 14
Test 1 10 21 13 43
Test 2 59 35 23 3
Test 3 63 33 16 6
Test 4 78 27 12 2
Experiment 6
Test 5 66 31 9 0
Test 6 96 24 3 0
Test 7 75 30 1 0
Test 8 15 16 18 36
Chapter 6
Discussion
The objective of this thesis was to review multiphase flow with special emphasis on
slug phenomena, followed by experiments to evaluate the use of a rotating device to
mechanically break down slugs in the gas-liquid interface and/or influence multiphase
flow in any significant and beneficial way. The results indicated varying impacts from
the turbine depending on flow configuration and placement. The largest impacts on the
flow behaviour was found in experiments 1, 2 and 6 as all these experiments studied the
impacts of the turbine when placed in the horizontal flow direction. The hydrodynamic
instability at the gas-liquid interface created slugs early in the flow loop, thus making it
possible to study the impacts of the turbine without the vertical section affecting the
flow.
The dimensions of the flow loop configurations limited the experimental study sub-
stantially. Terrain induced slugging required greater pressure build-ups than the di-
mensions in the experimental study would allow. This was caused by short lengths and
relatively low liquid flow rate. The horizontal flow loop was not as much limited by these
factors as hydrodynamic slugs developed over a broader range of parameters.
Terrain induces slugging and hydrodynamic slugging are very different in nature.
Terrain induced slugs occurs because of geometry allowing for liquid accumulation
at low spots and pressure build-ups while hydrodynamic slugs occurs because of flow
conditions and interaction between gas and liquid phase. Terrain induced slugs are
considered more harsh than hydrodynamic slugs because of large pressure and flow rate
fluctuations.
Terrain induced slugging did not occur in the flow loops containing bend sections
because of the dimensions and flow conditions. The turbine did not have any impact on
the flow because there was nothing to fix at the low points of the bend sections since gas
66 Discussion
was transported efficiently with the liquid flow as seen in Figure 5.6. Slug flow occurred
in the flow loops containing bend section, but the formation of the slugs took place at
other places of the flow loops rather than the base of the bend sections. It was discussed
briefly in Section 4.2 why terrain induced slugging did not occur and found to be caused
by the short length of the inclined sections not allowing for pressure build-ups. It was
mentioned that the lengths of the bend section should have been at least 12-13 m and
25 m using the Bøe criterion, but the assumptions for the calculations made the perfect
conditions for terrain induced slugging occurring and the length should probably been
even longer.
It was identified that the turbine was able to break down small gas bubbles ascending
over the turbine in the vertical direction as seen from snapshots in Figure 5.8, but was
not able to break down Taylor bubbles as seen from snapshots in Figure 5.9. When Taylor
bubbles ascended over the turbine, rotation stopped because of a vortex at the point
right above the propeller probably caused by the change of flow direction by the propeller.
The liquid velocity upwards was reduced and further accelerated downwards by the
gravitation. It was believed that this phenomenon stopped the device from rotating.
The horizontal flow loop with the test conditions resulted in hydrodynamic slugging.
These hydrodynamic slugs were caused by instabilities at the gas-liquid interface and the
slugs increased in size along the flow. The turbine altered the flow by increasing the slug
frequency and reducing the size of the slugs. From the slow motion footage in Figure 5.15,
it was seen that the liquid height at the position of the propeller increased which reduced
the height of the gas at the same position. Studying the Kelvin-Helmholtz instability
criterion of Equation 3.1 when the gas height is reduced, revealed that instability were
more likely to occur and create slugs, explaining the increase of frequency with the
turbine. In the experiments in this study, local effects were studied, but larger scale and
different test intervals might have led to other results because of interaction between
slugs along the flow. The amount of large slugs were drastically reduced, but Figure 5.14
illustrated that short distance from turbine to test interval might have caused these
results. It was, however, evident that under these experimental conditions that the
turbine reduced the size of the slugs over a short distance.
It was mentioned briefly in Section 4.3 that the device reduced the surface area by
approximately 37.5 %. This constriction of flow area would affect the production rate in
a negative way. From Figure 5.16, it was seen that cavitation happened as the already
created slug flowed through the turbine. Cavitation is associated with fast formation and
67
breakdown of gas bubbles due to Bernoulli effect, here created as the result of reducing
the flow area.
Slug lengths are associated with uncertainty because of random flow behaviour since
slugs interact with each other over large distances and might coalesce to larger slugs
as well as slugs might break down. Even at the small scale of the experiments in this
study, slugs grew and broke down in the test interval which is why large increments
were used for the size distributions. The flow loops in this study were short and only
local effects were studied. By studying Figure 5.14, it was seen that the number of
large slugs decreased significantly when the device was placed further away from the
inlet. The device placement very close to the test interval, so that slugs did not have
the opportunity to accumulate to larger slugs, might have caused this. It could also be
caused by the impacts of the device on the flow behaviour, but yet again not allowing
for accumulation of larger slugs before the test interval. It was, however, evident that
there was an impact from the device on this scale despite the large uncertainties of
these experiments. Uncertainty also related to the method used to gather the results as
impacts have been captured on film and further results from visual inspections of the
films.
Chapter 7
Conclusion
This study was set out to explore the possibilities of utilizing a turbine to mechanically
affect the slug flow behaviour in undulated horizontal wells and pipeliner-riser systems.
As terrain induced slugging did not occur in any of the flow loops, it was decided to
study the impacts from the turbine on hydrodynamic generated slugs. The nature of
hydrodynamic slugs are very different from the nature of terrain induced slugs. hydrody-
namic slugs are caused by flow instability and interaction between gas and liquid phase
whereas terrain induced slugs are created from liquid accumulation at low spots. This
means that the turbine is not expected to have the same outcome on terrain induced
slugging as the results in this study. New experiments should be conducted with a flow
loop where terrain induced slugging occurs to study the impacts of the turbine on this
flow.
• Slug flow occurred in the flow loops caused by hydrodynamic instability as a result
of flowing conditions, but terrain induced slugging did not occur in any of the flow
loops likely because to the dimensions.
• Based on the results in experiment 5, it was indicated that the impact of the turbine
in the vertical flow direction was small, probably because the turbine stopped
rotating when large Taylor bubbles ascended over the turbine and the liquid phase
velocity was reduced.
70 Conclusion
• Based on the results from experiments 3 and 4, there was no impact from the
turbine on the flow behaviour in the bend sections.
• Based on the results from experiments 1, 2 and 6 in the flow loop consisting of
horizontal flow sections, slug flow tended to occur at the same spot in the pipe
with the test conditions. It was believed that placing the turbine around this
region would yield results as the turbine could break down the slugs as they were
generated. It was seen that the frequency increased while the size of each individual
slug was reduced. It was found that slugs tended to form over the turbine as the
liquid height was locally increased and waves grew due to flow instabilities. Further,
it was seen that placement of the turbine closer to the test interval resulted in the
lowest number of with size l s > 15 D as the slugs did not have the opportunity to
grow in the short distance.
• First, To evaluate the use of the turbine for terrain induced slugging, the dimensions
must be scaled much larger so that liquid slugs can accumulate at low points. With
larger dimensions comes possibilities of studying placement of turbine in inclined
section to evaluate optimal positioning. As discussed in Section 3.4.3, the flow
in the downward flow direction have to be effectively stratified. Investigation on
whether or not the turbine could affect this criterion could be feasible.
• Identify the possibility of a new device to transport the gas in inclined section
down to the base point or modifying the device in this study to the extent that it is
used for transportation of gas in the downward inclined section.
References
Andreussi, P., Paglianti, A. and Silva, F. S. (1999), ‘Dispersed bubble flow in horizontal
pipes’, Chemical Engineering Science 54(8), 1101 – 1107.
Aziz, K. and Govier, G. W. (1972), The Flow of Complex Mixtures in Pipes, New York: Van
Nostrand Reinhold Company.
Bai, Y. and Bai, Q. (2012), Subsea Engineering Handbook, Gulf Publishing Company.
Baker, O. (1953), Design of pipelines for the simultaneous flow of oil and gas, in ‘Fall
Meeting of the Petroleum Branch of AIME’, Society of Petroleum Engineers.
Barnea, D., Shoham, O. and Taitel, Y. (1982), ‘Flow pattern transition for downward in-
clined two phase flow; horizontal to vertical’, Chemical Engineering Science 37(5), 735–
740.
Bøe, A. (1981), ‘Severe slugging characteristics; part i: flow regime for severe slugging; part
ii: Point model simulation study’, Selected Topics in Two-Phase Flow, NTH, Trondheim,
Norway .
Cadei, L., Montenegro, G., Scaramellini, S., Di Sarra, A., Cerri, P., Bianco, A., Masi, S. et al.
(2015), Slug catcher multiphase cfd modelling: Optimization and comparison with
industrial standards, in ‘Offshore Mediterranean Conference and Exhibition’, Offshore
Mediterranean Conference.
Corneliussen, S., Couput, J.-P., Dahl, E., Dykesteen, E., Frøysa, K.-E., Malde, E., Moestue,
H., Moksnes, P. O., Scheers, L. and Tunheim, H. (2005), ‘Handbook of multiphase flow
metering’, Norwegian Society for Oil and Gas Measurement (NFOGM), Revision 2.
Dahl, B. (2015a), ‘Katalog 2015 teknisk vvs’. Retrieved from
http://dahl.no/Documents/Kataloger/VVS/KATALOG_Teknisk_2015.pdf.
Dahl, B. (2015b), ‘Prisbok 2015 industri’. Retrieved from
http://www.dahl.no/Documents/Kataloger/Industri/Produktbok_Industri_2015.pdf.
Dukler, A. E. and Hubbard, M. G. (1975), ‘A model for gas-liquid slug flow in horizontal and
near horizontal tubes’, Industrial & Engineering Chemistry Fundamentals 14(4), 337–
347.
Enilari, B., Kara, F. et al. (2015), Slug flow and it’s mitigation techniques in the oil and gas
industry, in ‘SPE Nigeria Annual International Conference and Exhibition’, Society of
Petroleum Engineers.
Feesa, L. (2003), ‘Hydrodynamic slug size in multiphase flowlines’. Retrieved from
http://www.feesa.net./case-studies/.
72 References
Filip, A., Băltăreţu, F. and Damian, R.-M. (2014), ‘Comparison of two-phase pressure
drop models for condensing flows in horizontal tubes’, Mathematical Modelling in
Civil Engineering 10(4), 19–27.
Gokcal, B., Al-Sarkhi, A., Sarica, C., Alsafran, E. M. et al. (2009), Prediction of slug fre-
quency for high viscosity oils in horizontal pipes, in ‘SPE Annual Technical Conference
and Exhibition’, Society of Petroleum Engineers.
Gregory, G. and Scott, D. (1969), ‘Correlation of liquid slug velocity and frequency in
horizontal cocurrent gas-liquid slug flow’, AIChE Journal 15(6), 933–935.
Haheim, S., Gaillard, X. et al. (2009), A simplified subsea separation and pumping system,
in ‘SPE Annual Technical Conference and Exhibition’, Society of Petroleum Engineers.
Heywood, N. and Richardson, J. (1979), ‘Slug flow of air—water mixtures in a horizontal
pipe: Determination of liquid holdup by γ-ray absorption’, Chemical Engineering
Science 34(1), 17–30.
Hill, T. (1990), ‘Gas injection at riser base solves slugging flow problems’, Oil and Gas
Journal;(USA) 88(9).
Jahanshahi, E. (2013), ‘Control solutions for multiphase flow: Linear and nonlinear
approaches to anti-slug control’.
Jansen, F., Shoham, O. and Taitel, Y. (1996), ‘The elimination of severe slug-
ging—experiments and modeling’, International journal of multiphase flow
22(6), 1055–1072.
Jansen, F., Shoham, O. et al. (1994), Methods for eliminating pipeline-riser flow instabili-
ties, in ‘SPE Western Regional Meeting’, Society of Petroleum Engineers.
Li, S.-M. (2007), Mean-Field Free-Energy Lattice Boltzmann Method For Liquid-Vapor
Interfacial Flows, PhD thesis, Virginia Polytechnic Institue and State University.
Malekzadeh, R. (2012), Severe slugging in gas-liquid two-phase pipe flow, TU Delft, Delft
University of Technology.
Malekzadeh, R., Henkes, R. and Mudde, R. (2012), ‘Severe slugging in a long pipeline–
riser system: experiments and predictions’, International journal of multiphase flow
46, 9–21.
Malekzadeh, R. and Mudde, R. F. (2012), A modelling study of severe slugging in wellbore,
in ‘North Africa Technical Conference and Exhibition’, Society of Petroleum Engineers.
Mandhane, J., Gregory, G. and Aziz, K. (1974), ‘A flow pattern map for gas—liquid flow in
horizontal pipes’, International Journal of Multiphase Flow 1(4), 537–553.
Manolis, I., Mendes-Tatsis, M. and Hewitt, G. (1995), The effect of pressure on slug
frequency on two-phase horizontal flow, in ‘The 2nd conference on multiphase flow,
Kyoto, Japan’.
Norris, H. L. (2012), The use of a transient multiphase simulator to predict and suppress
flow instabilities in a horizontal shale oil well, in ‘SPE Annual Technical Conference
and Exhibition’, Society of Petroleum Engineers.
References 73
Obendrauf, W., Schrader, K., Al-Farsi, N., White, A. et al. (2006), Smart snake wells in
champion west-expected and unexpected benefits from smart completions, in ‘SPE
Asia Pacific Oil & Gas Conference and Exhibition’, Society of Petroleum Engineers.
Ogazi, A. I. (2011), Multiphase severe slug flow control, Cranfield University.
Øverland, A. M. and Ramstad, H. J. (2001), Yme marginal field, 12 km subsea gas lift
experience, in ‘SPE Annual Technical Conference and Exhibition’, Society of Petroleum
Engineers.
Pots, B. F., Bromilow, I. G., Konijn, M. J. et al. (1987), ‘Severe slug flow in offshore flow-
line/riser systems’, SPE Production Engineering 2(04), 319–324.
Experiments
A.1 Experiment 1
Table A.1: Results with turbine 120 cm from inlet resulting from a period of 4 minutes.
u S L = 0.20ms−1 & high gas rate.
A.2 Experiment 2
Table A.2: Results with turbine 195 cm from inlet resulting from a period of 4 minutes.
u S L = 0.20ms−1 & high gas rate.
A.3 Experiment 3
Table A.3: Results with turbine in bend section resulting from a period of 4 minutes.
u S L = 0.20ms−1 & high gas rate.
A.4 Experiment 4
Table A.4: Results with turbine in two bend sections resulting from a period of 4 minutes.
u S L = 0.20ms−1 & high gas rate.
A.5 Experiment 5
Table A.5: Results with turbine in vertical section resulting from a period of 4 minutes.
u S L = 0.20ms−1 & high gas rate.
A.6 Experiment 6
Table A.6: Results without turbine and turbine 80 cm from inlet resulting from a period
of 4 minutes. u S L = 0.20ms−1 & high gas rate.
Test #1 Test #2
Without turbine 80 cm from inlet
Table A.7: Results with turbine 100 cm and 120 cm from inlet resulting from a period of
4 minutes. u S L = 0.20ms−1 & high gas rate.
Test #3 Test #4
100 cm from inlet 120 cm from inlet
Table A.8: Results with turbine 140 cm and 160 cm from inlet resulting from a period of
4 minutes. u S L = 0.20ms−1 & high gas rate.
Test #5 Test #6
140 cm from inlet 160 cm from inlet
Table A.9: Results with turbine 200 cm from inlet and without turbine resulting from a
period of 4 minutes. u S L = 0.20ms−1 & high gas rate.
Test #7 Test #8
200 cm from inlet Without turbine